>Didn't know you could specify decimal with a `0t' prefix.  It's not in the
>manpage.  Could you add this stuff?

Err, it's in there. See "number" under EXPRESSIONS.

>Also looking at the manpage:
>
>    write[/bhl] address expression [expression ...]
>I think we need to add /L for 64-bit longs.

Sounds reasonable. Want to implement it? Shouldn't it be "q"
to be consistent with printf(3) syntax, though?

>(It would also be nice to be able to specify different address spaced
>for reads and writes so I could directly examine physical memory.  At
>the moment I have a machine extension to dump physical RAM, but for
>writing or devices I need to go to PROM.)

"Hmm." It'd also be nice to do reads and writes from io space
(I have a "machine {in|out}" implemention for the i386 sitting
on the back burner...).  Do you have some clear idea of how to do this
in a unified fashion?

>> "ps" could be modified to prin the pid in the current radix.
>
>That sounds good.  Or prefix decimal with `0t' so you don't have to guess
>and possibly kill the wrong PID.

Hmm, that sounds better. I think there's value in having ps show
pid's that match those shown by /usr/bin/ps.

>One other thing that bugs me is that if your current radix is 16 and you
>type in a hex number that starts with an `a'-`f' ddb will try to look it
>up as a symbol and print an error message.

Ah, yes. This is annoying. I end up just prefixxing them with 0x.

>I'd like it if it also tried converting it as hex before complaining.

Well, s/as hex/as the current radix/.

The problem with that is then we'd need a way to unambiguously
specify symbol names.  I guess we have that as "netbsd:alldevs".
I'll look into that.

Manuel wrote:

| Ok, I didn't know radix. So with $radix=0t10, 11 is really 11 (decimal) ?

0t explicitly means decimal (t==ten, x==hex).

--jhawk